Influence of fish-farming management on duck ... - Springer Link

2 downloads 0 Views 349KB Size Report
Nov 27, 2009 - Influence of fish-farming management on duck breeding in French fish pond systems. Joël Broyer • Clément Calenge. Received: 22 August ...
Hydrobiologia (2010) 637:173–185 DOI 10.1007/s10750-009-9994-3

PRIMARY RESEARCH PAPER

Influence of fish-farming management on duck breeding in French fish pond systems Joe¨l Broyer • Cle´ment Calenge

Received: 22 August 2007 / Revised: 20 October 2009 / Accepted: 13 November 2009 / Published online: 27 November 2009 Ó Springer Science+Business Media B.V. 2009

Abstract Fish ponds host a relatively important share of the breeding population of waterfowl in Europe. The objective of this study was to investigate the influence of fish-farming management on the distribution of dabbling and diving duck breeding, in 103 fishponds from four regions important for duck breeding in France. Duck breeding in fish ponds was apparently influenced by food resource and nesting site availability. Dabbling duck pair density was generally higher when invertebrates were abundant in macrophytes beds and when competition for food with carps Cyprinus carpio was potentially decreased by lower fish stock biomass. Diving duck pair abundance seemed to be negatively influenced by low invertebrates biomasses in pond sediment and by the absence of large reed beds at the edge of waterbodies. The difference between dabbling and diving ducks was also confirmed by a study of the variation in brood numbers standardized by pair numbers. This brood:pair ratio corresponded to a measure of nesting success and the possible attraction of fish ponds for some broods hatched in neighbouring waterbodies. Dabbling duck brood:pair ratio was

Handling editor: Steven Declerck J. Broyer (&)  C. Calenge Office National de la Chasse et de la Faune Sauvage, Direction des E´tudes et de la Recherche, Station de la Dombes, Montfort, 01330 Birieux, France e-mail: [email protected]

lower in one of the study regions where most of meadow areas surrounding fish ponds have been transformed into cereal crops in the past few decades. Diving duck brood:pair ratio was positively linked to pond fertilization due to fish farming. Fish farming, therefore, influenced duck breeding by an interaction between carp density and fertilization. We hypothesize that fertilization increases zooplankton density, thereby decreasing the competition between carps and waterfowl for benthic prey. Keywords Duck  Breeding  Fish pond  Fish-farming  Invertebrates  Nesting cover

Introduction Several studies have emphasized the positive consequences of absence or scarcity of fish for duck breeding conditions in lake ecosystems (Pehrsson, 1979; Hill et al., 1987; Parker, 1991; Desgranges & Gagnon, 1994; Mallory et al., 1994). Competition between duck and fish for invertebrates as a food resource could influence habitat selection by female ducks and affects brood home ranges and thus brood density (Eriksson, 1979; Anderson, 1981; Mc Eadie & Keast, 1982; Desgranges & Rodrigue, 1986). Fish can severely reduce invertebrate availability; for instance, after experimental fish removal from a gravel pit in Great Britain, chironomids and molluscs were three

123

174

times more abundant and tufted duck Aythya fuligula brood survival increased (Giles, 1994). It could thus be seen as a paradox that fish pond regions, where fish productivity is often very high, are among important breeding wetlands for ducks in Europe (Bukacinska et al., 1996; Musil, 1999; Lutz, 2001). Fish-farming in fish ponds may influence duck breeding: (i) through competition with fish for food resources, (ii) through consequences of pond fertilization or artificial food supply for carps Cyprinus carpio, (iii) through the destruction of shallow littoral areas with aquatic emergent vegetation as potential nesting sites. Grazing by carps, which are generally predominant in European fish ponds, might also affect duck habitat characteristics through their impact on water transparency and the development of macrophytes as well as benthic invertebrate and plankton communities (Spencer & King, 1984; Meijer et al., 1990; Wright & Phillips, 1992; Hanson & Butler, 1994; Lilie & Evrard, 1994; Bouffard & Hanson, 1997; Pokorny & Pechar, 2000). Indeed, ducks partly feed on invertebrates on their breeding grounds (Sugden, 1973; Swanson, 1984; Noyes & Jarvis, 1985; Jacobsen, 1991; Cox et al., 1998), which they find in the sediment of waterbodies or in submerged or floating macrophyte leaves (Krull, 1970; Danell & Sj}oberg, 1982). In general, diving ducks are more carnivorous than dabbling ducks (Cramp & Simmons, 1977). However, numerous studies on the feeding ecology of primarily herbivorous dabbling duck species have confirmed that they switch from plants and seeds to a diet dominated by animal matter during reproduction (Dirschl, 1969; Swanson & Serie, 1974; Serie & Swanson, 1976; Jacobsen, 1991). In North Dakota, the consumption of invertebrates increased in the diet of female pintail Anas acuta from 56% during prelaying to 77% during laying (Krapu, 1974). Swanson et al. (1979) found that invertebrates may constitute from 70% (in mallard Anas platyrhynchos) to 99% (in blue-winged teal Anas discors) of the food of laying hens in five species of dabbling ducks. Invertebrate abundance may thus regulate pond use by dabbling ducks in spring (Joyner, 1980). Ducklings themselves depend heavily on the consumption of invertebrates as a source of protein during their first weeks of life. During such periods, competition for food with fish may determine the breeding waterbody selection by birds (Winfield & Winfield, 1994; Petr,

123

Hydrobiologia (2010) 637:173–185

2000). Duck distribution may also be influenced by nest site availability. The existence of wide strips of littoral vegetation, often considered by fish-farmers as unproductive areas, may provide secure breeding conditions in increasing the distance between the nest and the habitat edge (Albrecht et al., 2006). Nest predation rates could thus be lower in large vegetation units (Krasowski & Nudds, 1986). In order to investigate the influence of fish-farming management on duck breeding in French fish ponds, this paper aims at answering four complementary questions: (1) does invertebrate availability during the egg-laying and duckling rearing periods have an influence on duck distribution among fish ponds? (2) Do fish-farming practices implemented to artificially increase food availability for carps (supplemental feeding, pond fertilization) have an impact on duck breeding? (3) Does nesting site availability influence the distribution of pairs and broods? (4) To what extent can fish stock biomass increase without interfering with duck breeding? Since their respective feeding behaviour and nesting site selection differ, we made separate analysis for dabbling ducks and diving ducks.

Methods Study areas The study was carried out in four of the most important duck breeding regions in France: Brenne in central France, Dombes and Forez in the east and Champagne in the north. Brenne and Dombes have more than 1,000 ponds each, Forez has 250 and in Champagne there are less than 100. In Champagne, they are much more scattered over the landscape. The distance in metre (mean(SE)) to the five nearest ponds was 225.7 (14.6) in Dombes, 240.5 (18.7) in Brenne, 242.6 (36.6) in Forez and 2,196.5 (450.6) in Champagne. The mean fish pond surface area varies from 7 ha in Forez to 10 ha in Dombes. Carps always form the bulk of the fish production. Supplemental carp feeding (barley, maize, soya) is provided in some ponds of Brenne and, more rarely, of Dombes. It is virtually never practised in Champagne and Forez. Ducks may in some circumstances also benefit from artificial feeding. For example, we saw several times in Brenne, pochard Aythya ferina adults and broods

Hydrobiologia (2010) 637:173–185

feeding on flour spread by automatic floating devices. Organic or mineral manure can be spread, in decreasing frequency from Brenne to Champagne and to Dombes. Fish ponds are not currently fertilized in Forez, although large amounts of phosphoric industrial residues have been added to the sediments from the 1930s to the late 1980s to increase plankton biomass (Robin, 1999). Zooplankton occurs then in remarkable densities, described as ‘plankton soup’ by Forez fish-farmers, with large species such as Daphnia magna. In Dombes, the zooplankton communities are more scattered and made up of smaller species (Daphnia longispina, Bosmina longirostris), not easily filtered by carp gills (Fanget, 1976; Vallod, 1984). European pochard and mallard are always the most abundant duck species breeding in these areas. Gadwall, Anas strepera, and tufted duck also breed in all the four study regions, whereas red-crested pochard, Netta rufina, only does so in Dombes and Forez. Shoveler, Anas clypeata, garganey, Anas querquedula, and teal, Anas crecca, are more sporadic. Over the last two decades, duck breeding populations have increased in Brenne, decreased in Dombes, and have been stable in Forez (Broyer, 2002). The trend in Champagne is unknown. In 2000 and 2001, the two larger regions were sampled, i.e. Brenne and Dombes. The main difference between the two regions was a massive loss of meadow habitat at the edge of fish ponds in Dombes after conversions to cereal crops since 1970 (Broyer, 2000). Additional data were collected in Forez in 2002 and in Champagne in 2003. The reason for extending the initial sampling to the other two fish pond areas were the high degree of eutrophication in Forez as a result of the phosphoric content of pond sediment (see above), and the more isolated situation of fish ponds in Champagne which limits the possibilities for breeding adults to feed in neighbouring waterbodies.

175

Fish pond selection We selected ponds from a regional set of 60 for which the duck population (pairs and broods) had already been counted the previous year. Of the initial 60-pond set, we selected the 10 ponds with the highest and the 10 ponds with the lowest brood densities in the preceding year with the objective to maximize the variation of duck abundance in samples. We had to discard some of them for which data were incomplete (Table 1). Duck distribution in fish ponds In all the selected ponds, adult ducks and broods were counted weekly with a telescope from mid-April to the end of July. Annually in each region, all the ponds were visited by the same observer throughout the study period. The behaviour and distribution (individuals alone, pairs or groups) of adults in the two first 10-day periods of May allowed us to assess the number of territorial pairs for each species. Age and number of ducklings in observed broods were recorded every week to determine the number of broods present in every fish pond. Broods were monitored until the age of 4 weeks (main growing period). A preliminary data examination indicated that the pair number was positively correlated with pond surface area and that pair density was negatively correlated with pond surface area. Therefore, pair numbers were divided by the square root of the pond area to allow comparison between ponds of different sizes (the transformed variable was no longer related to pond surface area). We cannot ascertain that territorial pairs actually nested in the ponds in which they were observed nor that all the observed broods were born there. We know examples of colour-marked females with ducklings that were successively reported in different waterbodies (ONCFS, unpublished).

Data collection Invertebrate biomass The following variables were studied in each selected pond: duck pair and brood density, invertebrate biomass in the sediment and in macrophyte leaves, the surface of emergent littoral vegetation and macrophyte beds, and pond management (fish stock, fertilization, artificial carp feeding) as described by interrogated fish-farmers.

Our purpose was to assess the biomass density of benthic invertebrates available in the sediment just before and in the early stage of egg-laying (first half of May), and in submerged or floating macrophyte leaves once macrophyte beds were fully grown (i.e. mid-June, at the onset of the duckling rearing period).

123

176

Hydrobiologia (2010) 637:173–185

Table 1 Regional characteristics of sampled fish ponds [for continuous variables = mean (SE)] Dombes (46°000 N 05°010 E)

Brenne (46°490 N 01°130 E)

Forez (45°360 N 04°030 E)

Champagne (48°430 N 04°350 E)

Total fishponds

[1,000

[1,000

250

\100

n Fishponds in samples

2000: 15 2001: 16

2000: 16 2001: 17

2002: 19

2003: 20

Total sampled area

750 ha

413 ha

253 ha

386 ha

Carp feeding frequency

25.8%

54.5%

0

0

Fertilization frequency

12.9%

54.5%

0

40.0%

Fish stock (kg/ha)

201 (15)

268 (23)

296 (55)

286 (33)

Shore vegetation (%)

5.8 (0.7)

13.3 (1.9)

13.8 (2.2)

10.2 (2.3)

Reed beds (ha)

0.8 (0.2)

1.2 (0.4)

1.4 (0.3)

2.4 (0.8)

Invertebrates in sediment (g/m2)

6.8 (1.9)

2.8 (0.5)

1.1 (0.2)

2.2 (0.4)

Invertebrates in macrophytes (g/l)

2.3 (0.4)

2.7 (0.4)

4.2 (0.8)

3.4 (0.8)

Dabbling duck pair density

2.0 (0.4)

1.6 (0.2)

2.2 (0.4)

1.0 (0.2)

Diving duck pair density

1.6 (0.3)

1.3 (0.2)

1.8 (0.3)

0.6 (0.1)

Dabbling duck brood:pair ratio

0.4 (0.1)

0.9 (0.2)

0.6 (0.1)

0.4 (0.1)

Diving duck brood:pair ratio

0.6 (0.1)

1.2 (0.3)

1.6 (0.4)

0.8 (0.2)

A 10-cm-thick layer of sediment (cf. Lafont, 1989) was collected in a 225-cm2 area with an Ekman dredge. For each pond, samples were taken in three different places at the upper extremity (opposite to the outlet) of the waterbody at a depth of less than 1 m: (i) in clear aquatic vegetation, (ii) at the interface between aquatic vegetation and open water i.e. nonvegetated areas, (iii) in open water. The three samples were pooled prior to invertebrate biomass measurements. Submerged macrophytes (mainly Potamogeton sp., Chara sp., Callitriche sp., Ranunculus peltatus, Najas marina and N. minor, Myriophyllum spicatum and Ceratophyllum demersum) were sampled by two operators with a (25 cm 9 25 cm) square landing net (0.1-mm mesh) to collect the plant parts in the upper 25 first cm of the water column, i.e. the most accessible to young ducks and dabbling ducks. Floating and submerged vegetation was collected in the landing net over a maximum distance of 1 m. The net was plunged vertically, very slowly in the heart of the selected macrophyte stands, swiftly pulled forward and extracted from the water in horizontal position with collected vegetation hanging inside. The latter was then cut with scissors so as to fall either inside or outside the net. At least three samples of the most representative aquatic vegetation were collected in each pond and pooled. Sample number

123

(3–5) varied according to macrophyte bed diversity (diverse dominant species or vegetation density). The collected vegetation volume was measured by water displacement in a test tube; the shifted volume of water after the introduction of plant material was directly used as a reference to compare invertebrate biomass density between different sites (in gram/litre of macrophytes). Invertebrates were separated from plants or sediment under a water flow and sorted by successive sieving (2-mm mesh, then a 0.5-mm one). According to Morin et al. (2004), a 1-mm sieve retains on average [90% of the invertebrate biomass. The invertebrates of the sediment were sub-sampled in a squared box with 49 compartments. Analysed compartments were selected randomly until at least 200 individuals were found. Their biomass (mg dry weight) was measured after desiccation for 24 h at 60°C. Invertebrate biomass density in sediments was expressed in g/m2. Aquatic vegetation Macrophyte abundance was assessed in June during the sample collection for invertebrate biomass study. For each pond, macrophyte bed coverage was classified into one of three classes (1 = \10% of total pond area, 2 = 10–30%, 3 = [30%) after examination

Hydrobiologia (2010) 637:173–185

from all round the pond perimeter and incursions into the waterbodies. The emergent littoral vegetation area was measured after the interpretation of aerial photographs. Only vegetation was taken into account whose presence was linked to the direct influence of the waterbody, i.e. growing in water or in wet adjacent grounds: Phragmites communis, Typha angustifolia or T. latifolia, Phalaris arundinacea, Carex sp., Scirpus lacustris, Juncus effusus and J. conglomeratus, Glyceria maxima. Surface area in hectares of total helophytes was transformed into percentages of the total fish pond area. We also estimated the surface area of reed beds i.e. the total area covered by high helophytes (Phragmites ? Typha ? Scirpus lacustris ? Glyceria maxima). Data analysis Fish ponds are complex ecosystems in which environmental variables, fish-farming practices and interactions likely to interfere with duck breeding are numerous and diverse. Our study, therefore, takes place in an exploratory context. The year and region effects being confounded for Forez (2002) and Champagne (2003), the analysis was conducted first with the data from Brenne and Dombes (2000, 2001). We then used the data from Forez and Champagne to compare the results obtained for Brenne and/or Dombes with the outcome of a similar analysis in the specific conditions found in the other two regions: high phosphorus content of pond sediment in Forez and pond isolation in Champagne. Our aim was to identify explanatory variables most strongly correlated with four response variables: (i) dabbling duck pair density, (ii) diving duck pair density, (iii) brood:pair ratio in dabbling ducks, (iv) brood:pair ratio in diving ducks. The brood:pair ratio, i.e. the number of different family groups (whatever the duckling number in each one) observed in each pond divided by the pair number, is a rough assessment of nesting success (the average number of broods per pair). This assessment could be biased by possible brood movements between neighbouring waterbodies. However, a preliminary exploration revealed that the number of broods and the number of pairs were closely related (Spearman r [ 0.6 in all the species). These four response variables are characterized by the presence of numerous outliers. We used a rank

177

transformation of the data to remove the effect of these outliers. This transformation lead to more robust and powerful results in case of such unstable distributions (Conover & Iman, 1981). Once the environmental variables affecting the responses were identified, we confirmed the detected patterns in the data by drawing graphs of the untransformed variables. We considered the following explanatory variables: region (R), year (Y), invertebrate biomass in pond sediment (IS) and in macrophyte beds (IM), fish stock biomass in kg/ha (FS), presence of carp feeding (CF), of fertilization (F), abundance of macrophyte (M), percentage of shore emergent vegetation (B), and size of the reed beds (Re). All continuous explanatory variables (IS, IM, B and Re) were characterized by highly asymmetrical distributions with a large number of outliers on both sides of their distribution. The treatment of outliers is problematic in applied statistics. They may cause serious difficulty to model fit, such as strong leverage. No transformation allowed us to remove their effect and we had no other choice than to transform all the continuous explanatory variables into binomial data: (i) lower than or equal to the median value of the variable in the sample, (ii) greater than the median value of the variable in the sample. Note that all the variables but one (M, which had three levels) had only two levels, which simplified the subsequent analyses. We used General linear model (GLM) and especially analysis of variance (ANOVA, which is a special case of GLM) as an exploratory tool to identify the environmental variables affecting response variables. For each response variable, we built an explanatory model predicting the response according to the effects of explanatory variables and their firstorder interactions. The main difficulty with GLM is the selection of explanatory variables to be included in the best model (Guisan & Zimmermann, 2000). Most existing methods have been developed for confirmatory studies and are not suitable for exploratory ones. For example, the use of information theoretical methods (e.g., Akaike Information Criterion) is not recommended (Burnham & Anderson, 1998), since this method may easily lead to the identification of spurious patterns (Anderson et al., 2001). Similarly, many authors have warned against the abuse of hypothesis testing in exploratory studies (Cohen, 1994; Nester, 1996; Cherry, 1998; Johnson, 1999).

123

178

In order to illustrate this point, we generated a completely random response variable i.e. randomly distributed ranks and we searched, among all possible explanatory variables and interactions, the apparently most important effect on this random response. We repeated this process 1000 times, and at each step, we computed the F-ratio associated with the largest ‘effect’ (ratio between the amount of variance explained by an effect and the residual variance), as well as the corresponding P-value. In 75% of cases, the F-ratio was higher than 5, and the associated P-value lower than 0.02. Hypothesis testing may, therefore, identify ‘statistically significant’ effects on a random response variable. We, therefore, did not use any hypothesis test, whether parametric or not, since they would not have been valid. However, we based our variable selection procedure on this simulation: for each response variable, we considered all the possible effects and interactions (when interactions between two variables were included in a model, we also included the main effects) using an ANOVA: we decided to include an effect in the model by considering the value of the F-ratios. Our simulation indicated that, when the response variable is completely random (i.e. not related to the explanatory variables), 95% of the F-ratios were lower than 11.5, and 90% were lower than 10. We included in the model the effects with F-ratios C10, to limit the risk of inclusion of spurious effects in the model. The use of a F-ratio threshold common to all the response and explanatory variables was possible because: (i) all the explanatory variables but one had the same degree of freedom (two levels = one degree of freedom), and (ii) all the response variables were rank-transformed and distributed on the same scale.

Hydrobiologia (2010) 637:173–185

density in macrophytes ([2.0 g/l) and low fish stock (\220 kg/ha) (Fig. 1). Diving duck pair density The interaction between the ‘region’ and ‘carp feeding’ variables had a very strong effect on diving duck pair density (F = 14.51). The interaction between invertebrate biomass density in the sediment and the size of reed beds also had a strong effect on this variable (F = 9.31). Although the F-ratio associated to this latter interaction was \10, we decided to include this effect in the model since: (i) the F-ratio was not very different from 10, (ii) the difference between the sum of squared residuals of the models with and without this effect was very high (= 2,324, while being very generally much lower than 1,500), and (iii) invertebrate biomass density in the sediment and reed beds are both of biological interest for diving ducks. A graphical exploration of this latter interaction (Fig. 2B) indicated that its effect was probably due to the difference in pair density between ponds with both small reed beds (\0.3 ha) and low invertebrate biomass density in the sediment (\2.7 g/m2) (lower pair density), and other ponds (higher

Results Dombes and Brenne Dabbling duck pair density The best model for dabbling duck pair density included only the interaction between invertebrate biomass density in macrophytes and fish stock density (F = 12.84). All other effects had a low F-ratio (F \ 6). Actually, the number of dabbling duck pairs was higher in ponds with high invertebrate biomass

123

Fig. 1 Variation of dabbling duck pair density (number/square root of pond area) in Brenne and Dombes according to the explanatory variables included in the model: fish stock density (low \220 kg/ha; high C220 kg/ha) and invertebrate biomass density in macrophytes

Hydrobiologia (2010) 637:173–185

179

supplied with carp feeding (Fig. 2A), but in Brenne conversely when ponds were supplied with carp feeding (Fig. 2B). Dabbling duck brood:pair ratio The brood:pair ratio in dabbling ducks was strongly related to the ‘region’ effect (F = 12.6) and all other effects were relatively minor in comparison (F \ 3). In dabbling ducks, the brood:pair ratio was higher in Brenne than in Dombes (Fig. 3A). Diving duck brood:pair ratio There was a very strong effect of the interaction between fertilization and the size of reed beds (F = 19.27). Diving duck brood:pair ratio was higher in fertilized ponds than in non fertilized ones, except in ponds with large reed beds (Fig. 3B). It is interesting to notice that invertebrate biomass density in macrophytes was also higher in fertilized ponds (Fig. 4). Champagne and Forez

Fig. 2 Boxplots showing the variation of diving duck pair density (number/square root of pond area) according to the explanatory variables included in the models: A carp feeding in Dombes, and B carp feeding and the interaction between invertebrate biomass density in the sediment (IS) and the size of reed beds (Re) in Brenne

pair density). All other effects exhibited a low proportion of explained variation. These results suggest that diving duck pairs were less abundant in ponds with small or no reed bed and low invertebrate biomass density in sediment. They were more abundant in Dombes when ponds were not

The data collected in Champagne and Forez were used to confirm the hypotheses established in Brenne and/or Dombes i.e. the effects of invertebrate biomass in macrophytes and fish stock biomass on dabbling ducks, of invertebrate biomass in sediment and the size of reed beds on diving ducks and the influence of fertilization on the brood:pair ratio. They were subsequently tested in the particular conditions of Forez (high phosphoric load of pond sediments) and Champagne (more isolated waterbodies) where carp feeding is unpractised. Champagne fish ponds had lower duck pair densities (Table 1). In particular, diving duck pairs were absent from a high number of ponds (30% in Champagne vs. 3–12% in the other regions). Very low invertebrate biomass density was found in the sediment of ponds without diving duck pairs (Fig. 5A) but diving ducks were present in all the ponds (n = 5) in which large reed beds were absent. The positive influence of high invertebrate biomass density in macrophytes and low fish stock on dabbling duck density was not confirmed there (mean dabbling duck pair density = 0.23, SE = 0.12, n = 3 vs. mean = 0.95, SE = 0.23, n = 12 in the other ponds).

123

180

Hydrobiologia (2010) 637:173–185

Fig. 4 Boxplots showing the variation of invertebrate biomass density in macrophytes in Brenne and Dombes (France), between fertilized and unfertilized ponds

in macrophytes (mean dabbling duck pair density = 1.77, SE = 0.58, n = 4 vs. 2.63, SE = 0.90, n = 8 in the other ponds).

Discussion

Fig. 3 A Boxplot showing the variation of dabbling duck brood:pair ratio between Brenne and Dombes; B variation of diving duck brood:pair ratio according to the explanatory variables included in the model: fertilization and reed bed size

In Champagne, the brood:pair ratio was higher in fertilized ponds (Fig. 5B) and, among fertilized ponds, higher in ponds with large reed beds (mean = 1.33, SE = 0.58, n = 3 vs. mean = 0.58, SE = 0.20, n = 5 ponds without large reed beds). Forez fish ponds differed from those in any other study region in having very low IS values. Despite this phenomenon, diving duck pair density and the brood:pair ratio were the highest in this region (Table 1). Large reed beds were present in all the sampled fish ponds. Moreover, dabbling duck pair densities were not higher in ponds with both low fish stock and high invertebrate biomass density

123

Our results stress the complexity of interactions between environmental variables and fish farming practices, with resulting difficulty to describe their respective influence on fish pond ecosystem. Food resources seemed to influence duck pair distribution in Brenne and Dombes. Dabbling duck pair density was higher in ponds where invertebrates were abundant in macrophyte beds and when competition with carps was potentially decreased by lower biomass. Diving duck pair density tended to be lower when invertebrate abundance was low in pond sediments; moreover, in ponds with supplemental carp feeding, diving duck densities were lower in Dombes but higher in Brenne. In both regions, carp feeding increased fish stock biomass (on average, ?28% in Dombes and ?25% in Brenne); however, 61% of fish ponds with carp feeding were also fertilized in Brenne, but only 25% in Dombes. It is also probable that accessibility for ducks to the food brought to carps may vary with the different methods implemented (flour floating on water surface or

Hydrobiologia (2010) 637:173–185

Fig. 5 Boxplots showing, in Champagne (France): A the variation in invertebrate biomass density of sediments between fish ponds with and without the presence of diving duck pairs; B the variation of the brood:pair ratio between fertilized and unfertilized ponds (diving and dabbling ducks pooled)

crushed grains thrown in deep water, spread throughout duck breeding time or only at the end of the cold season). These results were partly confirmed in Champagne where ponds are more isolated: invertebrate abundance also seemed to influence diving duck distribution since low biomass density was found in the sediment of ponds that were avoided by territorial pairs. In Mayenne (western France) where

181

waterbodies are scattered as well, Le Louarn & Birkan (2000) found low pochard pair frequency in fish ponds with lower chironomid density in the sediment in April and during the first half of May. In Forez however, low invertebrate biomass density generally found in pond sediment did not affect diving duck abundance. In Champagne, we could not confirm the relationship between dabbling duck distribution and the interaction between fish biomass and invertebrate abundance in macrophytes: lower duck densities (Table 1) could possibly decrease the competition between ducks themselves or the lack of choice resulting from the low pond density and number may lead dabbling ducks to select suboptimal habitat more often. Nesting site availability might also affect duck breeding. The presence of large reed beds seemed to compensate for low invertebrate density in pond sediment in Brenne and in Dombes: diving duck pair density was indeed higher when reed beds were present. For dabbling ducks, the difference in the brood:pair ratio observed between Brenne and Dombes could be explained by the massive conversion of meadow habitat surrounding fish ponds to cereal crops in Dombes since 1970 (Broyer, 2000). According to Greenwood et al. (1995), gadwall nesting success in Canada decreased by ca. 4% for every 10% increase in meadow conversion into arable land. Nesting success was then probably affected by nest site availability: the lower brood:pair ratio in Dombes could be accounted for by the shortage in meadow habitat close to fish ponds in which mallards or gadwalls usually nest, spacing their nests over large areas which could make them more difficult to be detected by predators (G}oransson et al., 1975; Owen & Black, 1990). Our results, thus, suggest that nesting site availability may influence pair distribution in diving ducks or nesting success in dabbling ducks. For dabbling ducks, the opportunities for potential nesting in nearby grounds seem to outweigh the influence of fish pond management on breeding success. The paradoxical impacts of fish-farming practices are well illustrated by the correlation between diving duck pair density and supplemental carp feeding which was negative in Dombes where most fish ponds were not fertilized, and positive in Brenne where most fish ponds were fertilized. Diving duck brood:pair ratio was higher in fertilized ponds of

123

182

Brenne and Dombes, except when large reed beds were present. This surprising influence of reed beds could be explained by fish-farming management: compared to unfertilized fish ponds, fish stock biomass increased on average by 58% in fertilized ponds without large reed beds and by 90% in fertilized ponds with large reed beds, while 67% of the former and only 25% of the latter received supplemental carp feeding. Competition with carps for invertebrates could then possibly hamper diving duck breeding in fertilized ponds with large reed beds. In Champagne, however, the brood:pair ratio was higher in fertilized ponds and the highest in fertilized ponds with large reed beds. Similarly, the highest brood:pair ratios were found in Forez whereas ponds had sediment with high phosphoric content and large reed beds. These results are consistent with the positive correlation between bird number or biomass and habitat trophic status as observed in lake ecosystems (Hoyer & Canfield, 1994). However, the mechanism of how fertilization could influence duck breeding remains unclear. Usually, fertilization stimulates plankton development (Lewkowicz & Lewkowicz, 1976; Barbe et al., 2000; Schlumberger, 2002). In fish ponds, carps feed on plankton first, when large species (size C500 lm) temporarily proliferate in May, June and July (Wunder, 1938; Grygiereck, 1979; Vallod, 1984; Robin, 1999), and shift to benthos after the pond’s depletion in late summer (Fanget, 1976). Their competition with ducks for benthos might then possibly be limited during the months when ducks use mostly benthic prey. In a macrophyte poor lake, Giles et al. (1990) found that consumption of cladoceran plankton by roaches, Rutilus rutilus, minimized diet overlap between roach and waterfowl. Moreover, ducklings may possibly feed on dense flocks of large zooplankton species (Bartonek & Hickey, 1969) and, thus, could be less dependent on benthic invertebrates. This could explain the apparent absence of correlation between duck and invertebrate abundance in Forez. Theoretically, invertebrate abundance could also depend on the abundance of planktonic prey. We found, indeed in Forez, on average the highest invertebrate biomass density in macrophytes while in Dombes and Brenne invertebrate abundance in macrophytes was higher in fertilized ponds. Habitat difference between the study regions might then explain the observed difference in trophic

123

Hydrobiologia (2010) 637:173–185

relationships. Similarly in Champagne, diving ducks were absent in ponds where invertebrate biomass density in the sediment was low, potentially due to the difficulty to find extra food resources in nearby waterbodies. The influence of fertilization, which enhanced fish stock density (mean = 328.7 kg/ha, SE = 22.8 vs. 194.3, SE = 11.7 in unfertilized ponds), was then paradoxical. The impact of carps on submerged macrophytes has been experimentally studied in Camargue (France): the decrease of aquatic plant biomass was assessed at 20, 35 and 45% of their dry weight, when carp biomass amounted respectively to 400–500, 600 and 700 kg/ha (Crivelli, 1983). Moreover, above an upper threshold of nutrient concentration, plant communities in shallow waterbodies shift from a clear-water state dominated by submerged macrophytes to a more eutrophic state dominated by phytoplankton with limited light available to macrophytes (Scheffer, 1990; Scheffer et al., 1993). In this study, however, macrophyte abundance in fish ponds was not found to be negatively affected by fish stocks C400 kg/ha (macrophyte coverage[30% in 63.2% of 19 ponds vs. 41.1% of 56 in which fish stock was \400 kg/ha). Fish-farming influence appeared, therefore, to be the result of an interaction between carp biomass and fertilization (or carp feeding, when artificially supplied food is also accessible to ducks in the breeding period). Increases in the breeding duck population in central Russia in the 1960s, in Belarus in the 1980s, or in the wintering pochard population in eastern Germany from the 1960s to the end of the 1980s, corresponded to quantitative progress in fish-farming (Rutschke & Liebherr, 1995; Sukhanova, 1996; Kozulin et al., 2002). After restrictions in carp feeding, pochard and tufted duck populations have undergone, respectively, 55 and 73% decreases in Belarus between 1990 and 2001 (Kozulin et al., 2002). This suggests that artificial food either was used by carps and by ducks, or decreased the competition between carps and ducks for invertebrates. In the Czech Republic, the ever-increasing productivity of fish ponds since the 1950s did not hamper pochard expansion until the 1970s. Later, however, greater gains in fish biomass were achieved to the detriment of duck populations, both adults (Pykal & Janda, 1994) and broods (Albrecht et al., 2000). Musil et al. (2001) consider that ‘suitable fish pond conditions for

Hydrobiologia (2010) 637:173–185

waterfowl breeding may be characterized as fish stock density being less than 400 kg/ha’. In conclusion: (1) duck pair distribution was linked with invertebrate biomass available in macrophytes to dabbling ducks and in pond sediments to diving ducks, (2) artificially increasing food availability for carps through supplemental feeding or fertilization may positively influence duck breeding, probably as far as macrophyte development is not dramatically disturbed, (3) nesting site availability seemed to influence diving duck pair distribution and dabbling duck nesting results and (4) fish stock biomass in French fish ponds did not appear to limit duck breeding in general, but competition with carps seemed to interfere with dabbling duck pair distribution. Acknowledgements The authors thank colleagues/ fieldworkers who contributed to the data collection: F. Bourguemestre, L. He´rault, L. Huguet (Brenne), M. Benmergui, J. Y. Fournier, L. Curtet, C. Chimenton (Dombes), J. B. Mouronval, A. Canny (Champagne), G. Chavas, D. Delayre, G. Pluvier, R. Delaroque (Forez) and V. Pereira (Institut Supe´rieur d’Agriculture Rhoˆne-Alpes) for invertebrate collection and analysis, C. Bernard for fish farmer questioning in Champagne. This study was funded by the following institutes: ONCFS, Conseil Ge´ne´ral de la Loire, Fe´de´ration des chasseurs de l’Indre. We also warmly thank Catherine Carter (ONCFS) for the improvement of the English writing.

References Albrecht, T., P. Musil & J. Cepak, 2000. Habitat selection of waterfowl broods on intensively managed fishponds in Czech Republic. Sylvia 36(suppl.): 18. Albrecht, T., D. Horak, J. Kreisinger, K. Weidinger, P. Klvana & T. C. Michot, 2006. Factors determining pochard nest predation along a wetland gradient. Journal of Wildlife Management 70: 784–791. Anderson, G., 1981. Influence of fish on waterfowl in lakes. Anser 20: 21–34. Anderson, D., K. Burnham, W. Gould & S. Cherry, 2001. Concerns about find effects that are actually spurious. Wildlife Society Bulletin 29: 311–316. Barbe, J., O. Schlumberger & N. Bouretz, 2000. Evaluation de la production piscicole potentielle des e´tangs. Ingenierie – EAT 11: 49–62 (in French). Bartonek, J. C. & J. J. Hickey, 1969. Selective feeding by juvenile diving ducks in summer. The Auk 86: 443–457. Bouffard, S. H. & M. A. Hanson, 1997. Fish in waterfowl marshes: waterfowls manager’s perspective. Wildlife Society Bulletin 25: 146–157. Broyer, J., 2000. La Dombes, espace d’e´quilibre ou simple substrat pour la culture ce´re´alie`re. Courrier de l’Environnement de l’INRA 40: 63–65.

183 Broyer, J., 2002. Re´sultats compare´s de la reproduction des anatide´s dans trois principales re´gions de nidification de France: la Dombes, la Brenne, le Forez (in French with English summary: Comparative breeding success of Anatidae in three important breeding areas: the Dombes, the Brenne and the Forez). Alauda 70: 377–386. Bukacinska, N., D. Bukacinski, J. P. Cygan, K. A. Dobrowolski & W. Kaczmarek, 1996. The importance of fishponds to waterfowl in Poland. Acta Hydrobiologica 37: 57–73. Burnham, K. P. & D. R. Anderson, 1998. Model selection and inference. Springer, Berlin. Cherry, S., 1998. Statistical tests in publications of the wildlife society. Wildlife Society Bulletin 26: 947–953. Cohen, J., 1994. The earth is round (p \ .05). American Psychologist 49: 997–1003. Conover, W. & R. Iman, 1981. Rank transformations as a bridge between parametric and nonparametric statistics. The American Statistician 35: 124–129. Cox, R. M., M. A. Hanson, C. Roy, N. H. Euliss, D. H. Johnson & M. G. Butler, 1998. Mallard duckling growth and survival in relation to aquatic invertebrates. Journal of Wildlife Management 62: 124–133. Cramp, S. & K. E. L. Simmons, 1977. Handbook of the Birds of Europe, the Middle East and North Africa. Volume I, Ostrich to Ducks. Oxford University Press, Oxford: 722 pp. Crivelli, A. J., 1983. The destruction of aquatic vegetation by carps. Hydrobiologia 106: 37–41. Danell, K. & K. Sj} oberg, 1982. Successional patterns of plant invertebrates and ducks in a man-made lake. Journal of Applied Ecology 9: 345–409. Desgranges, J. L. & C. Gagnon, 1994. Duckling response to changes in the trophic web of acidified lakes. Hydrobiologia 280: 207–221. Desgranges, J. L. & J. Rodrigue, 1986. Influence of acidity and competition with fish on the development of ducklings in Quebec. Water, Air and Soil Pollution 30: 743–750. Dirschl, H. J., 1969. Foods of lesser scarp and blue-winged teal in the Saskatchewan River Delta. Journal of Wildlife Management 33: 77–87. Eriksson, M. O. G., 1979. Competition between freshwater fish and goldeneyes for common prey. Oecologia 41: 99–107. Fanget, R., 1976. Contribution a` l’e´cologie des e´tangs piscicoles de la Dombes: sur le re´gime alimentaire de la Carpe a` miroirs Cyprinus carpio. The`se, Universite´ Lyon I (in French). Giles, N., 1994. Tufted duck (Aythya fuligula) habitat use and brood survival increases after fish removal from gravel pit lakes. Hydrobiologia 279/280: 587–592. Giles, N., M. Street & R. M. Wright, 1990. Diet composition and prey preference of tench, Tinca tinca, common bream, Abramis brama, perch, Perca fluviatilis and roach, Rutilus rutilus, in two contrasting gravel pit lakes: potential trophic overlap with waterfowl. Journal of Fish Biology 37: 545–957. G} oransson, G., J. Karlsson, S. G. Nilsson & S. Ulfstrand, 1975. Predation on bird nests in relation to antipredator aggression and nest density: an experimental study. Oı¨kos 26: 117–120. Greenwood, R. J., A. B. Sargeant, D. H. Johnson, L. M. Cowardin & T. L. Schaffer, 1995. Factors associated with

123

184 duck nest success in the Prairie Pothole Region of Canada. Wildlife Monograph 128: 1–57. Grygiereck, E., 1979. Plankton as an ecological indication of the influence of the farming measures on ponds biocenosis. Polish Ecological Studies 5: 77–140. Guisan, A. & N. E. Zimmermann, 2000. Predictive habitat distribution models in ecology. Ecological Modeling 135: 147–186. Hanson, M. A. & M. G. Butler, 1994. Response of plankton, turbidity and macrophytes to biomanipulation in a shallow prairie lake. Canadian Journal of Fisheries and Aquatic Sciences 51: 1180–1188. Hill, D., R. Wright & M. Street, 1987. Survival of mallard ducklings and competition with fish for invertebrates on a flooded gravel quarry in England. Ibis 129: 159–167. Hoyer, V. M. & D. E. Canfield, 1994. Bird abundance and species richness on Florida lakes: influence of trophic status, lake morphology, and aquatic macrophytes. Hydrobiologia 297/280: 107–119. Jacobsen, O. W., 1991. Feeding behaviour of breeding Wigeon Anas penelope in relation to seasonal emergence and swarming behaviour of chironomids. Ardea 79: 409–418. Johnson, D., 1999. The insignificance of statistical significance testing. Journal of Wildlife Management 63: 763–772. Joyner, D. E., 1980. Influence of invertebrates on pond selection by ducks in Ontario. Journal of Wildlife Management 44: 700–705. Kozulin, A., O. Ostrovski, F. Tanneberger, O. Pareiko & V. Yurko, 2002. Waterbirds of large fishponds of Belarus. OMPO Special Publication, Vilnius. Krapu, G. L., 1974. Foods of breeding pintails in North Dakota. Journal of Wildlife Management 38: 408–417. Krasowski, T. P. & T. D. Nudds, 1986. Microhabitat structure of nest sites and nesting success of diving ducks. Journal of Wildlife Management 50: 203–208. Krull, J. N., 1970. Aquatic plant-macroinvertebrate associations and waterfowl. Journal of Wildlife Management 34: 707–718. Lafont, M., 1989. Contribution a` la gestion des eaux continentales: utilisation des oligoche`tes comme descripteurs de l’e´tat biologique et du degre´ de pollution des eaux et des se´diments. The`se, Universite´ Lyon I (in French). Le Louarn, H. & M. Birkan, 2000. The impact of physicochemical and biological characteristics of ponds on the nesting of the common pochard in Mayenne. Game and Wildlife Science 17: 129–146. Lewkowicz, M. & S. Lewkowicz, 1976. Organic and inorganic nutrient enrichment and the living conditions of carp fry in first rearing ponds. Physico-chemical factors and zooplankton. Acta Hydrobiologica 18: 235–257. Lilie, R. A. & J. O. Evrard, 1994. Influence of macroinvertebrates and macrophytes on waterfowl utilization of wetlands in the Prairies Pothole Region of north-western Wisconsin. Hydrobiologia 279/280: 235–246. Lutz, M., 2001. Les e´tangs de pisciculture en Europe estcentrale. Typologie des syste`mes d’exploitation et impacts des modalite´s de gestion sur l’avifaune (in French with English summary: Fishponds in Eastern Europe. Typology of fish-farming systems and effects of pond management on aquatic vegetation and nesting birds). The`se, Universite´ Strasbourg I.

123

Hydrobiologia (2010) 637:173–185 Mallory, M. L., D. K. Mc Nichol & P. J. Weatherhead, 1994. Habitat quality and reproductive effort of common goldeneyes nesting near Sudbury, Canada. Journal of Wildlife Management 58: 552–560. Mc Eadie, J. A. & A. Keast, 1982. Do goldeneye and perch compete for food? Oecologia 55: 225–230. Meijer, M.-L., E. H. R. R. Lammens, A. J. P. Raat, M. P. Grimm & H. Hosper, 1990. Impact of cyprinids on zooplankton and algae in ten drainable ponds. Hydrobiologia 191: 275–284. Morin, A., J. Stephenson, J. Skrike & A. G. Solimini, 2004. Sieve retention probabilities of stream benthic invertebrates. Journal of the North American Benthological Society 23: 383–391. Musil, P., 1999. Monitoring of water bird breeding populations in the Czech Republic (1988–1997). Vogelwelt 119: 253–256. Musil, P., J. Cepak, K. Hudec & J. Zarybnicky, 2001. The long-term trends in the breeding waterfowl populations in the Czech Republic. OMPO/Institute of Applied Ecology, Kostelec nad Cernymi lesy. Nester, M., 1996. An applied statistician’s creed. Applied Statistics 45: 401–410. Noyes, J. H. & R. L. Jarvis, 1985. Diet and nutrition of breeding female redhead and canvasback ducks in Nevada. Journal of Wildlife Management 49: 203–211. Owen, M. & J. M. Black, 1990. Waterfowl Ecology. Chapman & Hall, New York, USA. Parker, G. R., 1991. Survival of juvenile American black ducks on a managed wetland in New Brunswick. Journal of Wildlife Management 55: 466–470. Pehrsson, O., 1979. Feeding behaviour, feeding of diving ducks breeding at lake Myvatn, Iceland. Ornis Fennica 48: 77–92. Petr, T., 2000. Interactions between fish and aquatic macrophytes in inland waters. A review. FAO Fisheries Technical Paper No 396, Rome. Pokorny, J. & L. Pechar, 2000. Development of fishpond ecosystems in the Czech Republic: role of management and nutrient input (Limnological review). Sylvia 36: 8–14. Pykal, J. & J. Janda, 1994. Pocetnost vodnich ptaku na rybnicich ve vztahu k rybnicnimu hospodareni. Sylvia 30: 3–11 (in Czech). Robin, J., 1999. Dynamique saisonnie`re du phytoplancton en e´tang de pisciculture de la plaine du Forez (Loire). Essai de controˆle des cyanobacte´ries. The`se, Universite´ Lyon I (in French). Rutschke, E. & H. Liebherr, 1995. Bestand und Bestandentwicklung einiger Wildentenarten in Ostdeutschland. Beitra¨ge zur Jagd und Wildforschung 20: 261–273 (in German). Scheffer, M., 1990. Multiplicity of stable states in freshwater systems. Hydrobiologia 200/201: 475–486. Scheffer, M., S. H. Hosper, M.-I. Meijer, B. Moss & E. Jeppesen, 1993. Alternative equilibria in shallow lakes. Trends in Ecology and Evolution 8: 275–279. Schlumberger, O., 2002. Me´mento de pisciculture d’e´tangs. Cemagref, Paris, France (in French). Serie, J. R. & G. A. Swanson, 1976. Feeding ecology of breeding gadwalls on saline wetlands. Journal of Wildlife Management 40: 69–81.

Hydrobiologia (2010) 637:173–185 Spencer, C. & D. King, 1984. Role of fish in regulation of plant and animal communities in eutrophic ponds. Canadian Journal of Fisheries and Aquatic Sciences 41: 1851–1855. Sugden, L. G., 1973. Feeding ecology of pintail, gadwall, American widgeon, and lesser scaup duckling in southern Alberta. Canadian Wildlife Service Report Series 24: 45 pp. Sukhanova, O. V., 1996. Nesting ecology of the tufted duck and the pochard in Central Russia. Game and Wildlife 13: 709–722. Swanson, G. A., 1984. Invertebrates consumed by dabbling ducks (Anatinae) on the breeding grounds. Journal of the Minnesota Academy of Science 50: 3–40. Swanson, G. A. & J. R. Serie, 1974. Feeding ecology of breeding blue-winged teals. Journal of Wildlife Management 38: 396–407. Swanson, G. A., G. L. Krapu & J. R. Serie, 1979. Foods of laying female dabbling ducks on the breeding grounds. In

185 Bookhout, T. A. (ed.), Proceedings of the 1977 Symposium, The Wildlife Society, Madison, WI. Vallod, D., 1984. Suivis physico-chimiques et biologiques d’e´tangs de la Dombes. Application a` la gestion piscicole. The`se, Universite´ Lyon I (in French). Winfield, D. K. & I. J. Winfield, 1994. Possible competitive interactions between overwintering tufted duck Aythya fuligula and fish populations of lough Neagh, Northern Ireland: evidence from diet studies. Hydrobiologia 279/ 280: 377–386. Wright, R. M. & V. E. Phillips, 1992. Changes in the aquatic vegetation of two gravel pit lakes after reducing the fish population density. Aquatic Botany 43: 43–49. Wunder, W., 1938. La nourriture naturelle de la Carpe dans les e´tangs et les moyens d’en de´velopper la production. Bulletin Franc¸ais de Pisciculture 112: 121–129 (in French).

123